Avaliable Docker Images

MK
Last updated 7 days ago

Name

Description

Included Python libraries

Other packages

CHAINER

chainer-5.0.0-cpu-py36

Chainer image with Intel ideep4py installed.

Chainer (5.0.0)

-

chainer-5.0.0-gpu-py35-cuda9.1-openmpi2

Chainer image with CUDA-aware OpenMPI for distributed training.

Chainer (5.0.0), ChainerMN (1.3.1)

CUDA 9.1, cuDNN 7.1, OpenMPI 2.1.2

HOROVOD

horovod-0.15.0-cpu-py36-tf1.11.0

Horovod image with CPU-only OpenMPI and TensorFlow 1.11.0.

Horovod (0.15.0), TensorFlow (1.11.0)

OpenMPI (3.1.3)

horovod-0.15.2-gpu-py35-tf1.12.0-cuda9.0

Horovod image with GPU-aware OpenMPI and TensorFlow 1.12.0.

Horovod (0.15.2), TensorFlow (1.12.0)

CUDA 9.0, OpenMPI (3.1.2)

horovod-0.15.1-gpu-py35-tf1.12.0-cuda10.0

Image from NVIDIA GPU Cloud with Horovod 0.15.1, TensorFlow 1.12.0, and Python 3.5.

Horovod (0.15.1), Numba, Pandas, Scikit-learn, Scipy, TensorFlow (1.12.0rc2)

CUDA 10.0, cuBLAS 10.0, cuDNN 7.4, NCCL 2.3.7, OpenMPI 3.1.2

horovod-0.15.0-cpu-py36-pytorch0.4.1

Horovod image with CPU-only OpenMPI and PyTorch 0.4.1.

Horovod (0.15.1), Torch (0.4.1), Torchvision (0.2.1)

OpenMPI (3.1.3)

horovod-0.15.1-gpu-py35-pytorch0.4.1-cuda9.0

Horovod image with GPU-aware OpenMPI and PyTorch 0.4.1.

Horovod (0.15.2), Torch (0.4.1), Torchvision (0.2.1)

CUDA 9.0, OpenMPI (3.1.2)

OPENMPI

openmpi-3.1.3-cpu-py36

Ubuntu image with OpenMPI. Install other libraries using apt-get or pip.

-

OpenMPI (3.1.3)

openmpi-3.1.3-gpu-py36-cuda9.0

Ubuntu image with OpenMPI and NVIDIA CUDA 9.0. Install other libraries using apt-get or pip.

-

CUDA 9.0, cuDNN 7.0, OpenMPI (3.1.3)

openmpi-3.1.3-gpu-py36-cuda9.2

Ubuntu image with OpenMPI and NVIDIA CUDA 9.2. Install other libraries using apt-get or pip.

-

CUDA 9.2, cuDNN 7.1, OpenMPI (3.1.3)

PYTORCH

pytorch-0.4.1-cpu-py36

PyTorch 0.4.1 image with Python 3.6. Single instance only.

Pillow, Scipy, Torch (0.4.1), Torchvision (0.2.1)

Conda

pytorch-0.4.1-gpu-py36-cuda9.0

PyTorch 0.4.1 image with Python 3.6 and CUDA from Docker Hub. Single instance only.

Pillow, Scipy, Torch (0.4.1), Torchvision (0.2.1)

Conda, CUDA 9.0, cuDNN 7

pytorch-latest-cpu-py36

PyTorch image (installed from master branch) with Python 3.6. Single instance only.

Pillow, Scipy, Torch (1.0.0a0+78d594f), Torchvision (0.2.1)

Conda

pytorch-latest-gpu-py36-cuda9.0

PyTorch image (installed from master branch) with Python 3.6 and CUDA. Single instance only.

Pillow, Scipy, Torch (1.0.0a0+78d594f), Torchvision (0.2.1)

Conda, CUDA 9.0, cuDNN 7

pytorch-nightly-gpu-py36-cuda9.2

PyTorch nightly image with Python 3.6 from Docker Hub. Single instance only.

Torch-nightly (1.0.0.dev)

Conda, CUDA 9.2, CuDNN 7

TENSORFLOW

tensorflow-1.4.0-cpu-py27

TensorFlow 1.4.0 community image with Python 2.7 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.4.0)

-

tensorflow-1.4.0-gpu-py27-cuda8.0

TensorFlow-GPU 1.4.0 community image with Python 2.7 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.4.0)

CUDA 8.0, cuDNN 6.0

tensorflow-1.4.0-gpu-py27-cuda9.0

TensorFlow 1.4.0 image with Python 2.7 from NVIDIA GPU Cloud.

Horovod (0.11.3), TensorFlow (1.4.0)

CUDA 9.0, cuBLAS 9.0, cuDNN 7.1, NCCL 2.1.2, OpenMPI 3.0.0

tensorflow-1.4.0-cpu-py35

TensorFlow 1.4.0 community image with Python 3.5 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.4.0)

-

tensorflow-1.4.0-gpu-py35-cuda8.0

TensorFlow GPU 1.4.0 community image with Python 3.5 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.4.0)

CUDA 8.0, cuDNN 6.0

tensorflow-1.4.0-gpu-py35-cuda9.0

TensorFlow 1.4.0 image with Python 3.5 from NVIDIA GPU Cloud.

Horovod (0.11.3), TensorFlow (1.4.0)

CUDA 9.0, cuBLAS 9.0, cuDNN 7.1, NCCL 2.1.2, OpenMPI 3.0.0

tensorflow-1.8.0-cpu-py27

TensorFlow 1.8.0 community image with Python 2.7 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.8.0)

-

tensorflow-1.8.0-gpu-py27-cuda9.0

TensorFlow-GPU 1.8.0 community image with Python 2.7 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.8.0)

CUDA 9.0, cuDNN NCCL

tensorflow-1.8.0-gpu-py27-nvcr-cuda9.0

TensorFlow 1.8.0 image with Python 2.7 from NVIDIA GPU Cloud.

Horovod (0.12.1), TensorFlow (1.8.0)

CUDA 9.2, cuBLAS 9.0, cuDNN 7.1, NCCL 2.2.13, OpenMPI 3.0.0

tensorflow-1.8.0-cpu-py35

TensorFlow 1.8.0 community image with Python 3.5 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.8.0)

-

tensorflow-1.8.0-gpu-py35-cuda9.0

TensorFlow GPU 1.8.0 community image with Python 3.5 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.8.0)

CUDA 9.0, cuDNN NCCL

tensorflow-1.8.0-gpu-py35-cuda9.2

TensorFlow 1.8.0 image with Python 3.5 from NVIDIA GPU Cloud.

Horovod (0.12.1), TensorFlow (1.8.0)

CUDA 9.2, cuBLAS 9.0, cuDNN 7.1, NCCL 2.2.13, OpenMPI 3.0.0

tensorflow-1.11.0-cpu-py27

TensorFlow 1.11.0 community image with Python 2.7 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.11.0)

-

tensorflow-1.11.0-gpu-py27-cuda9.0

TensorFlow-GPU 1.11.0 community image with Python 2.7 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.11.0)

CUDA 9.0, cuDNN, NCCL

tensorflow-1.12.0-gpu-py27-cuda10.0

TensorFlow 1.12.0 image with Python 2.7 from NVIDIA GPU Cloud.

Horovod (0.15.1), TensorFlow (1.12.0-rc2)

CUDA 10.0, cuBLAS 10.0, cuDNN 7.4, NCCL 2.3.7, OpenMPI 3.1.2

tensorflow-1.11.0-cpu-py35

TensorFlow 1.11.0 community image with Python 3.5 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.11.0)

-

tensorflow-1.11.0-gpu-py35-cuda9.0

TensorFlow GPU 1.11.0 community image with Python 3.5 from Docker Hub.

Pandas, Pillow, Scikit-learn, Scipy, TensorFlow (1.11.0)

CUDA 9.0, cuDNN, NCCL

tensorflow-1.12.0-gpu-py35-cuda10.0

TensorFlow 1.12.0 image with Python 3.5 from NVIDIA GPU Cloud.

Horovod (0.15.1), TensorFlow (1.12.0-rc2)

CUDA 10.0, cuBLAS 10.0, cuDNN 7.4, NCCL 2.3.7, OpenMPI 3.1.2

XGBOOST

xgboost-0.81-cpu-py36-openmpi3

XGBoost image with OpenMPI and DMLC-Core, capable of running distributed using MPI back-end.

XGBoost (0.81)

OpenMPI 3.1.3